A MULTIPLE SINGLE-PASS HEURISTIC ALGORITHM FOR THE STOCHASTIC ASSEMBLY LINE REBALANCING PROBLEM

Assembly line re-balancing is a problem frequently tackled by companies, as continuous changes in product features and volume demand caused by the volatility of today’s markets produce assembly tasks redefinition and line cycle time fluctuations. Hence, managers have to adapt the balancing of their lines to accomplish with the new conditions, while trying to keep to a bare minimum increases in completion costs and in costs related to changes in tasks assignment. In particular, modifications in line balancing impact on operators training, equipment switching and moving, along with quality assurance. The stochastic assembly line re-balancing problem basically consists in a multi-objective problem where two objectives, total expected completion cost of the new line and similarity between the new and the existing line, have to be jointly optimized. In this paper, a multiple single-pass heuristic algorithm is consequently developed with the aim to find the most complete set of dominant solutions representing the Pareto front of the problem. Multiple single-pass procedures iterate the execution of single-pass algorithms, in order to generate a set of solutions, rather than to create a unique purpose. Given such a set, the best-performing solutions, in accordance with the multi-objective nature of the problem, are presented to the line designer, who selects the final assembly line balancing considering also additive factors that can be hardly inserted in a mathematical approach (i.e. simplicity of learning re-assigned tasks, time requested for the re-allocation of tools necessary for executing re-assigned tasks, experience requested for maintenance of tools necessary for executing re-assigned tasks). By means of a wide experimentation including comparisons with a multiobjective genetic algorithm, the behaviour of the proposed methodology is set and optimized.